Christophe Lyon <christophe.l...@arm.com> writes: > Hi Andrea, > > Minor comments below: > > On 4/28/23 13:29, Andrea Corallo via Gcc-patches wrote: >> Hi all, >> this patch fixes the vstrwq* MVE instrinsics failing to emit the >> correct sequence of instruction due to a missing predicates. Also the > nit: you have a typo, should be "predicate" > >> immediate range is fixed to be multiples of 2 up between [-252, 252]. > > Out of curiosity, which tests were affected by this error in the > immediate range?
Hi Christophe, no special reason, just because the test is autogenerated and we use this same pattern for all the intrinsics (the vast majority of which do return a non void type). The test indeed compiles fine so no problem there. BR Andrea